摘要：视觉语言模型正日益被用于驾驶场景理解，但其在全球南方交通结构较弱场景下的防御性推理能力仍不明确。我们提出 BD-HazardVLM，这是一个诊断性基准，包含 500 张孟加拉国道路场景图像，其中 375 张来自 RSUD20K，125 张来自 TFP-BD。该基准针对风险存在性、主要风险类别、风险时机、严重程度、防御性动作、可见证据，以及幻觉陷阱和西方道路偏见陷阱进行了标注。我们采用统一的零样本任务定义和结构化输出模式，对 Qwen2.5-VL-7B、InternVL3-8B 和 MiniCPM-V-4.6 进行了评估。二元风险存在性准确率介于 88.0% 至 89.2%，与 89.0% 的多数类基线基本相当。细粒度推理能力则显著更弱：类别准确率始终低于 40%，风险时机准确率介于 1.6% 至 12.4%，防御性动作的精确匹配率低于 2%。尽管 InternVL3-8B 以 61.2% 获得最高的严重程度准确率，但其在 58 个高严重度案例上的召回率为零。Qwen 出现严重的类别模式坍塌，而所有模型均表现出较弱的潜在风险预判能力，以及接近于零的自主陷阱识别能力。对于所评估的模型而言，这些发现表明，粗粒度风险识别并不意味着具备可靠的细粒度防御性驾驶推理能力，并进一步说明有必要开展地域多样化的安全评估。

Abstract:Vision-language models are increasingly used for driving-scene understanding, yet their defensive reasoning under weakly structured Global South traffic remains unclear. We introduce BD-HazardVLM, a diagnostic benchmark of 500 Bangladesh road-scene images, comprising 375 from RSUD20K and 125 from TFP-BD, annotated for hazard presence, primary hazard category, risk timing, severity, defensive actions, visible evidence, and hallucination- and Western-road-bias traps. We evaluate Qwen2.5-VL-7B, InternVL3-8B, and MiniCPM-V-4.6 using a shared zero-shot task definition and structured output schema. Binary hazard-presence accuracy ranges from 88.0 percent to 89.2 percent, essentially matching the 89.0 percent majority-class baseline. Fine-grained reasoning is substantially weaker: category accuracy remains below 40 percent, risk-timing accuracy ranges from 1.6 percent to 12.4 percent, and exact defensive-action matching remains below 2 percent. Although InternVL3-8B achieves the highest severity accuracy at 61.2 percent, it has zero recall on the 58 high-severity cases. Qwen exhibits severe category mode collapse, while all models show weak latent-hazard anticipation and near-zero self-reported trap recognition. For the evaluated models, these findings show that coarse hazard recognition does not imply reliable fine-grained defensive-driving reasoning and motivate geographically diverse safety evaluation.